Why did some firms perform better in the global financial crisis?

Abstract We explore what firm and macroeconomic factors assisted Chinese firms to resist the global financial crisis. We find that firms with higher top ten shareholder ratios or firms that are older exhibited saliently higher performance during the crisis, but performed poorly during the non-crisis period. Firm size has a notably negative impact on firm performance. Firms audited by the Big Four accounting firms have a significantly negative correlation with performance. During the crisis, stock markets became less efficient in incorporating firm-specific information into stock prices, signifying that the determinants of firm performance vary across non-crisis and crisis periods.


Introduction
Global equity markets dropped more than 56% and there was a reduction in equity value of more than $29 trillion in the recent global crisis (Bartram & Bodnar, 2009;Chen, Chen, & Lee, 2014a;Lee, Lin, & Zeng, 2016). Despite the overall negative impact of the global financial crisis, some companies profited during the market turbulence (Shakina & Barajas, 2014). This raises a fundamental question: what are the firm and macroeconomic determinants of firms' performance during crisis periods? In responding to the economic situation, governments and firm management adopted various policies or strategies to minimise the severity of the situation and to better cope with the challenges of the severe economic environment. Thus, performance of the firm depends on not only its ownership structure, financial features, and disclosure quality, but also external environments (period and macroeconomic elements).
Numerous studies have attempted to pinpoint some key variables that influence firm performance. For instance, Thomsen and Pedersen (2000) demonstrate that ownership structure is an important factor in firm performance, while others find no statistically significant relation between ownership structure and firm performance (Demsetz & Villalonga, 2001;Iannotta, Nocera, & Sironi, 2007). Some researchers argue that firm performance is KEYWORDS Global financial crisis; firm performance; contingency theory; panel data; china dependent on the extent of country or macroeconomic variables (Clarke, Cull, & Kisunko, 2012;Cutler, Poterba, & Summers, 1991). In particular, the recent literature has expanded the levels of research factors; for example, Dietrich and Wanzenried (2011) examined the effect of bank-specific, industry-specific, and macroeconomic determinants on bank profitability before and during the global crisis. Clarke et al. (2012) investigated how country and firm characteristics impact firms' financial constraints and their likelihood of survival during the early phase of the recent global crisis. To date, there is scant existing literature that explores the substantial determinants of non-financial firms and the impact of macroeconomic characteristics on firm performance during non-crisis and crisis periods (see Table 1). Robertson and Chetty (2000) note that, as a result of environmental changes, firms that have survived are expected to use experience and knowledge to discern information and hence make more informed decisions about critical variables. They suggest the contingency approach is a mid-range theory, because it holds the middle ground between two extreme positions. Therefore, this study investigates what firm and macroeconomic factors affected firm performance during the recent global financial crisis and non-crisis periods by adopting the contingency theory as a framework.
Building upon the works of Beltratti and Stulz (2012) and Dietrich and Wanzenried (2011), the present study contributes to related studies as follows. First, while an increasing number of scholars view a country's macroeconomic characteristics as imperative elements for firm performance (e.g., see Cutler et al., 1991;Hartmann, Kempa, & Pierdzioch, 2008;Mulder, Perrelli, & Rocha, 2012), there is little research discussing how firm and macroeconomic factors affect firm performance. Second, while several studies try to point out some important variables that influence firm performance during a crisis, there is scant existing literature that compares whether firm and macroeconomic characteristics leading to better firm performance are altered during diverse periods (i.e., during crisis and non-crisis periods). The exceptions appear to be Byström, Worasinchai, and Chongsithipol (2005), who reject the hypothesis that the default risks of firms are systematic before, during, and after the 1997 Asian crisis. By analysing the full time period and two sub-periods, we are able to see what the solid and temporary determinants are for better-performing firms during diverse time periods.
Third, most research on firm performance focuses on banks or several countries. For example, Dietrich and Wanzenried (2011) find that state-owned banks are more profitable than privately owned banks. Jin, Kanagaretnam, and Lobo (2011), Aebi, Sabato, and Schmid (2012), and Akhigbe, Madura, and Marciniak (2012) study the factors influencing U.S. bank performance during the financial crisis. Mitton (2002) and Gupta, Krishnamurti, and Tourani-Rad (2013) use several countries to explore whether corporate governance factors impact firm performance. Related studies have focused on a single country, such as Kim and Lee (2003), Baek, Kang, and Suh Park (2004), and Kang, Lee, and Na (2010) exploring South Korean firms, and Aebi et al. (2012) and Akhigbe et al. (2012) examining U.S. firms. To the best of our knowledge, only Liu, Uchida, and Yang (2012) highlight the specialty of Chinese state-owned enterprises (SOEs), in which these SOEs performed poorly before the global financial crisis, yet performed better during the crisis. The empirical evidence covering the determinants of Chinese firm performance has seldom been touched upon.
Due to the astonishing expansion and fast growth of China's stock markets, the management of large funds and international institutions have recognised the profit-making  Dietrich and Wanzenried (2011-2009 Examine how bank-specific characteristics, and industry-specific and macroeconomic factors affect the profitability of banks.
372 commercial banks in switzerland operationally efficient banks are more profitable than banks that are less operationally efficient. state-owned banks are more profitable than privately owned banks. 398 firms from five asian countries saliently better stock price performance is associated with firms that had indicators of higher disclosure quality, and with firms that had higher outside ownership concentration.
Firm kim and Lee (2003Lee ( ) 1997Lee ( -1998 Examine whether agency problems explain the performance of korean firms during the asian crisis and did agency problems explain the performance differently for chaebol vs. non-chaebol firms.
590 korean firms corporate governance matters in determining the role of agency problems. the explanatory power of agency problems is different for chaebol firms, compared with non-chaebol firms.
Firm Baek et al. (2004Baek et al. ( ) 1997Baek et al. ( -1998  opportunities and are weighting more of their investment portfolios in these stock markets (Chen, Lee, Lee, & Huang, 2016b;Lee, Chen, & Hsieh, 2013;Lao & Singh, 2011). Nevertheless, recent developments since the onset of the global financial crisis also show that China is not autonomous (Fidrmuc & Korhonen, 2010). We therefore employ annual data from 1,452 Chinese firms during the period 1999 to 2010 to investigate the determinants of firm and macroeconomic factors. The methodology of generalised method of moments (GMM) for panel data analyses is employed here to extract consistent and efficient estimates of the impact of firm and macroeconomic factors on firm performance.
We find that firms having higher top ten shareholder ratios or firms that are older performed poorly during the non-crisis period, yet performed better in the crisis period. For both crisis and non-crisis periods, firm size has pronouncedly negative impacts on firm performance, signifying that small firms could be more efficient. Firms audited by the Big Four accounting firms have a significantly negative correlation with their performance, signifying that the strict requirements held up by the Big Four accounting firms might lower firm performance. Leverage and audit opinions that are worse than unqualified audit opinions have salient impacts on firm performance during the non-crisis period rather than during the crisis period, implying that stock markets during a crisis become less efficient in incorporating firm-specific information into stock prices. Considering the macroeconomic factors, the consumer confidence index and macro-level exports have significant impacts on firm performance during all years. Our results reveal that the determinants of firm performance vary across non-crisis and crisis periods.
The remainder of the paper is organised as follows. Section 2 presents the literature review and variable selections. Section 3 provides the methodology and data. Section 4 presents the empirical results and discussion. Section 5 concludes the paper.

Literature review and determinants of firm performance
According to Bartram and Bodnar (2009), global equity markets originally had a market capitalisation of more than $51 trillion, but by the end of February 2009, global equity market capitalisation stood at just over $22 trillion, a drop of more than 56% and a reduction in equity value of more than $29 trillion. The loss in wealth to equity holders is equivalent in value to about 50% of global GDP for 2007. Rose and Spiegel (2012) also pinpoint that the severity of the global financial crisis and the crisis causes differ across countries. Bowman, Chan, and Comer (2010) find that the reactions of global equity markets to the 1997 Asian crisis were driven by macroeconomic fundamentals and countries with a weak economy as measured by poor macroeconomic fundamentals. This study thus explores the fast-growing but modestly researched Chinese markets to find the similarities and dissimilarities among better-performing firms across crisis and non-crisis periods. This section discusses the related literature and scrutinises likely applicable variables employed in the empirical analysis. See Table 2 for a summary of the variables used.

Dependent variables: firm performance
We use three alternative measures of firm performance. Following Aebi et al. (2012), we take two measures of firm performance: return on assets (ROA), defined as the firms' net income divided by total assets, and return on equity (ROE), defined as the firms' net income divided by the book value of equity. For robustness checks, this study follows Gomez-Mejia (1992) and Mashayekhi and Bazaz (2008) to construct earnings per share (EPS) in Renminbi (RMB) as a measure of firm performance. Research studies categorise EPS as a profitability measure, ROA as an operating performance measure, and ROE as a financial performance measure (Gomez-Mejia, 1992;Hart & Ahuja, 1996). Bartram and Bodnar (2009) mention that when the global financial crisis drove down stock markets, weaknesses in firms manifested themselves in a number of dimensions that left economies exposed when conditions began to turn even worse. During crises, Stone (2000)  suggests that specific firm elements lead to sharp falls in investment. Following the early work of Mitton (2002), a number of recent studies have attempted to identify some of the major determinants of firm performance. The empirical studies focus their analyses either on corporate governance evidence (e.g., Baek et al., 2004;Gupta et al., 2013;Liu et al., 2012) or financial firms (Akhigbe et al., 2012;Dietrich & Wanzenried, 2011;Jin et al., 2011). For instance, some scholars present that corporate governance is an important factor in firm value (La Porta, Lopez-de-Silanes, Shleifer, & Vishny, 2000;Lemmon & Lins, 2003;Mitton, 2002), while others find that firm performance is not related to governance (Cai, Qian, & Liu, 2009;Gupta et al., 2013). The empirical results of these above-mentioned studies present huge differences in their variables, time periods, investigated environments, and macroeconomic characteristics. This present study categorises the determinants of firm performance selected as follows. Disclosure quality. Lu, Liao, and Yang (2007) find that corporate value and information disclosure are correlated with a positive coefficient. Ongena, Smith, and Michalsen (2003) indicate that firm-specific information asymmetries could prevent some firms from accessing funds from outside sources. Mitton (2002) documents a significantly better stock return performance in firms that offered higher disclosure quality and greater transparency during the Asian crisis. Iatridis (2011) notes that the Big Four accounting firms tend to report higher quality disclosures (i.e., less earnings management), and firms that display high-quality accounting disclosure generally exhibit a larger size and higher profitability. Miihkinen (2012) shows that firm size is one of the important drivers of quality. Coad, Segarra, and Teruel (2013) argue that older firms are able to better convert slow growth into subsequent profits, but firm performance deteriorates with age. Forbes (2002) finds that larger firms often have worse performance than smaller firms after depreciation. Dodd, Dopuch, Holthausen, and Leftwich (1984) indicate that many firms experience negative abnormal performance prior to the release of qualified opinions, because there is increasing demand for greater transparency to be incorporated in their financial statements. In addition, the absence of research upon disclosure quality in Chinese firms motivates us to include four variables to proxy the level of information asymmetry: firms audited by the Big Four accounting firms, audit opinions, 1 the size of a firm in terms of the logarithm of sales, and a firm's age measured by the number of years since the public listed date. 2 Financial structure. Claessens, Djankov, and Nenova (2000) use the degree of financial leverage (debt over equity) and cash flow of a firm to reflect the corporate financial structure. Beltratti and Stulz (2012) find that better banks had less leverage before the global crisis. In contrast, Webber (2001) shows that firms with high leverage ratios grow fast, but they are vulnerable to external shocks. Mulder et al. (2012) state that the more a firm's financing is leveraged, the more likely its assets' value will severely erode its balance sheet when confronting a shock. They also indicate that the availability of assets can offset liquidity risks, which are captured by current ratios. Mulder et al. (2012) show that solvent firms run into problems if they lack liquid assets to meet obligations. Vuolteenaho (2002) decomposes firms' stock returns and finds that they are driven primarily by cash flow news. Dechow (1994) finds that the ability of cash flows to measure firm performance improves relative to earnings when the measurement interval is lengthened. We include cash flow per share, current ratio, and leverage ratio in this study due to their importance. Ding, Chua, and Fetherston (2005) indicate that value (growth) stocks can be purchased for prices that are low (high) relative to the companies' estimated underlying values such as price-earnings (P/E) and price-book (P/B). Basu (1983) and Jaffe, Keim, and Westerfield (1989) find positive abnormal returns associated with stocks with low P/E or P/B values. Bauman, Conover, and Miller (1998) reveal that P/E and P/B have a stable, negative relationship with stock returns, while P/E has originally been considered as an indicator of transitory earnings (Molodovsky, 1953), future earnings (Cragg & Malkiel, 1982), or risk (Ball, 1978). Danielson and Dowdell (2001) illustrate that a firm's P/B and P/E ratios can predict the future cash flow pattern earned by a firm. Lam (2002) find that P/B and P/E can capture the cross-sectional variation in average returns. Early in 1988, Fama and French documented the predictive power of dividend yields for stock returns. However, Chung and Lee (1998) report that dividends do not explain price movements. Chen and de Bondt (2004) reveal that dividend yield captures style-related trends in equity returns. Following the extant studies, this study employs P/E, P/B, and dividend yield ratio as the proxies of financial structure.

Independent variables: firm-level determinants of firm performance
Ownership structure. This study characterises ownership structure by top ten shareholders ratio, public share ratio, employees' share ratio, foreign institutions' share ratio, domestic institutions' share ratio, and state share ratio. Thomsen and Pedersen (2000) find a positive effect of ownership concentration on shareholder value and profitability. They further support that the identity of large owners -family, banks, institutional investors, government, and other companies -has important implications for corporate performance. However, Demsetz and Villalonga (2001) show no statistically significant relation between ownership structure and firm performance. After controlling for bank characteristics and country and time effects, Iannotta et al. (2007) conclude that ownership concentration does not significantly affect a bank's profitability.
A distinctive feature of the Chinese corporate ownership structure is the existence of state control (Liu et al., 2012). Sun and Tong (2003) and Gunasekarage, Hess, and Hu (2007) find a negative impact of state ownership on firm value, because the expropriation of minority shareholders by the state exists during normal economic states. Qi, Wu, and Zhang (2000) indicate that firm performance increases with the degree of relative dominance of legal person shares over state shares. However, Yu (2013) reveals that a higher level of state ownership is superior to a dispersed ownership structure due to the benefits of government support and political connections.
As for ownership concentration, Xu and Wang (1999) find that a positive correlation exists between performance and ownership concentration, and the performance is positively and significantly correlated with the fraction of legal person shares, rather than state shares. Admati, Pfleiderer, and Zechner (1994) show that large shareholders mitigate the free rider problem of monitoring a management team, thereby reducing the agency costs. In contrast, La Porta, Lopez-De-Silanes, and Shleifer (1999) argue that the existence of large shareholders with high stakes may be harmful to the firm, as the controlling shareholders' interests may not align with those of small shareholders. Bozec and Laurin (2008) indicate that firm performance is lower when large shareholders have both the incentives and the opportunity to expropriate minority shareholders. Few studies discuss the relationship between the top ten shareholder ratio and firm performance, and therefore we include the top ten shareholder ratio among the independent variables. Chidambaran and John (2000) show that institutional investors can convey private information obtained from management to other shareholders. Gillan and Starks (2003) show that institutional investors increase the liquidity of the markets and price informativeness and bring about a better monitoring of firms. In addition, Ferreira and Matos (2008) find that higher ownership by foreign and independent institutions has a significantly positive impact on firm valuation, unlike domestic ownership. Schuppli and Bohl (2010) prove that foreign institutions have a stabilising effect on Chinese stock markets. Baek et al. (2004) pinpoint that firms with larger foreign ownerships experienced a smaller drop in share value during the Asian crisis. This study hence separates institutional investors into foreign and domestic institutional investors to check the different impacts on firm performance.
As to public shareholders, Dyck and Zingales (2004) find that large shareholders presumably maximise their utility and thus need not align with public shareholder value. Morck and Yeung (2010) pinpoint that public shareholders' wealth is most directly tied to the firm's economic efficiency. Mitton (2002) demonstrates that firms with higher outside ownership concentration experienced better stock price performance during the Asian crisis. As for employees' shareholder ratio, Rayton (2003) document the existence of a link between the average pay of employees and the performance of the firm. de Bussy and Suprawan (2012) present that employee orientation contributes more to corporate financial performance than orientation towards any other individual primary stakeholder group. Thus, this study considers the variables that are used by the above-mentioned literature as follows: top ten shareholders ratio, public share ratio, employees' share ratio, foreign institutions' share ratio, domestic institutions' share ratio, and state share ratio. Cutler et al. (1991) establish that macroeconomic variables explain about one-third of the variance in stock returns. Ehrmann, Fratzscher, and Mehl (2009) show that countries with a stronger fiscal position prior to the recent global financial crisis were hit less severely during the crisis period, possibly due to the ability to conduct countercyclical fiscal policies more effectively. Dietrich and Wanzenried (2011) include bank-specific characteristics and macroeconomic factors in studying the determinants of profitability for banks in Switzerland before and during the global financial crisis. Thus, our non-financial firm performance determinants include macroeconomic factors, which have seldom been considered in previous studies. Corredor, Ferrer, and Santamaria (2013) find that the consumer confidence index captures investors' perceptions of the economic factors involved household spending and saving tendency, and has a positive influence on the volatility premium. Chen (2011) provides evidence for an asymmetric linkage between stock returns and consumer confidence: the impacts of returns on confidence are larger in bear markets. Bowman et al. (2010) mention that high values of M2 growth represent economic strength and have a positive relationship with returns. Baharumshah, Mohd, and Yol (2009) reveal that a stable long-term relationship exists between broad money supply (M2) and stock prices, as well as stock prices having a significant substitute (positive) effect on long-run M2 demand. Amiti and Weinstein (2011) pinpoint that a striking feature of many financial crises is the collapse of exports.

Macroeconomic-level determinants of firm performance
The producer price index (PPI) is a closely watched indicator of macroeconomic performance and a direct indicator of the purchasing power of money in various types of transactions. 3 Patton (2012) shows that monitoring PPI is important, because it influences stock performance. Sheikh et al. (2011) demonstrate that PPI is an important factor affecting profitability and an indicator of an increase in the interest rate which is considered for fighting against inflation. Freund and Pierola (2012) find that export expansion offers a path to robust income growth, especially when domestic demand lags, and that exports can sustain income growth. Given the importance of the above-mentioned macroeconomic variables, we include the consumer confidence index, M2 growth rate (GM2R), PPI, and exports (GEXR) as macroeconomic variables. 4

Crisis variable
This research focus is on the impact from different levels of features on a firm's performance during a crisis. However, this study needs to control other attributes that could potentially affect the performance of non-financial firms during the crisis. They are the global financial crisis effect and prior performance of the firms. 5 Following Grammatikos and Vermeulen (2012), we divide the period of the analysis into non-crisis (1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006) and the crisis periods (2007)(2008)(2009)(2010). In the all-period model, we set a crisis period dummy if the year is 2007-2010. Our inclusion of the control adds a level of comfort to our assertion that the interpretations of the significant coefficients on our reported variables are not due to correlations with omitted variables.

Empirical model
The dynamic panel approach offers advantages to ordinary least squares (OLS) and efficient to examine the firm factor, macroeconomic factors, and performance link using panel procedures. First, estimation using panel data allows us to explore the time-series nature of the relationship between firm factor, macroeconomic factors, and firm performances. Second, in a pure cross-firm instrumental variable regression, any unobserved firm-specific effect becomes part of the error term, which may bias the coefficient estimates, such as potential endogeneity of all explanatory variables. Thus, our panel procedures control for firm-specific effects. In addition, given the dynamic natures of the models, the OLS estimation methods produce biased and inconsistent estimates (Baltagi, 2001). The GMM technique for panel data analyses, proposed by Arellano and Bond (1991) and then further developed by Blundell and Bond (1998), is employed here to extract consistent and efficient estimates of the impact of firm and macroeconomic factors on firm performance, to control for endogeneity problems of sample firms, and to estimate individual firm's performance persistence. 6 This study focuses on the possible problem of endogeneity in firm performance estimation by using the two-step GMM estimator for our dynamic panel estimation, that is able to deal with the biases and inconsistencies of the estimates. 7 Further, concerning the instruments, we employ the Sargan statistic, which tests the over-identifying restrictions.
To explore the firm, period, and macroeconomic characteristics during the global crisis in driving the performance of firms' stock returns, this study follows García-Herrero, Gavilá, andSantabárbara (2009), andDietrich andWanzenried (2011), and uses a panel GMM model given by Equation (1): (1) where K it is the performance of each firm i at time t, with i=1,…, N and t=1,…T; P is the crisis period control variable; it are the firm-specific variables; η t are macroeconomic explanatory variables; and ɛ it is the disturbance. Coefficient α is the impact of crisis. A larger α means a greater influence of a crisis on firm performance, and a salient negative sign is expected.
Firm performances show a tendency to influence the next period, implying informational opacity and/or sensitivity to regional/macroeconomic shocks to the extent that they are serially correlated (Berger, Bonime, Covitz, & Hancock, 2011). Thus, this study specifies a dynamic panel data model by including a lagged dependent variable among the regressors, i.e., K i,t-1 is the one-period lagged performance and β ' is the speed of adjustment to equilibrium. A value of β ' between 1 and 0 indicates persistence of performance, but it will finally return to the normal level. A value close to 0 presents a firm that is fairly competitive, while a value close to 1 implies a less competitive structure.

Data
Since our interest is in understanding how firm factors, macroeconomic factors, and firm performances link differences during the non-crisis and crisis periods, we start from 1999 to 2006 as non-crisis period and the 2007-2010 as crisis periods. It is helpful to have the non-crisis data series start at the finish of Asian financial crisis (1997-1998), year 1999, 8 and end before 2007/01/01, as the unsustainable state of the U.S. subprime mortgage market was becoming ever more apparent in the second half of 2007 (Chor & Manova, 2012). In addition, Grammatikos and Vermeulen (2012), Shahrokhi (2011), and Choi (2013) Table 3 presents the summary statistics of the variables used. We take the variables of finance (cash flow per share (CF), current ratio (CUR), leverage ratio (LEV), price-to-earnings ratio (P/E), price-to-book ratio (P/B), dividend yield (DIV), return on assets (ROA), return on equity (ROE), earnings per share (EPS)), share structure (state share ratio (STATE), domestic institutions' share ratio (DOME), foreign institutions' share ratio (FORI), employees' share ratio (EMPL), public share ratio (PUB), and top ten shareholder ratio (TOP)), disclosure quality (CPA audited (CPA firm (CPA) size (logarithm of sales, SIZE), firm age (AGE), and audit opinion (OPION)), and Chinese macroeconomic data (consumer confidence index (CONF), producer price index (PPI), value of exports (GEXR), and annual growth rate of money supply (GM2R)) from https://www.cninfo.com.cn databases. To more closely examine the nature of better-performing firms during economic shocks, the annual data used in this study cover 1,452 publicly-traded non-financial, non-OTC firms that have a listed date before 2007 so as to cover before and during the global crisis periods. This paper follows the China Securities Regulatory Commission categorisation that categorises industries. 10 The major industry is manufacturing, followed by information technology. In addition, the sample effective year we use is ranged from 15 to 18 years over the sample period. On average, the firms in our sample have a ROE of 6.84%, ROA of 4.41%, and an EPS of 0.32 (in RMB) over the entire period from 1999 to 2010. The difference between mean and median indicates that large profitability differences exist among the firms in the sample. Public shares and state shares constitute the two largest ownership structures in our sample firms. Furthermore, this study winsorises the independent variables at the 1st and 99th percentiles to control for outliers. Table 4 provides the matrix of Pearson correlation coefficients. The correlation coefficients are usually small, except for GEXR having significantly negative correlations with the top ten shareholder ratio and public share ratio, as well as a negative correlation with the state share ratio. Table 5 summarises the GMM results for our main profitability measure ROA. The first two columns report the results when including all 12 years in our sample. In order to explore the determinants of firm performance during the crisis and non-crisis periods, we further split up two subsample periods from total sample period: the results of columns three and four refer to the non-crisis period (1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006) and columns five and six report the estimates for the years of the main financial crisis, namely 2007-2010. In order to identify the stability of the coefficients and their significance, we first include only the firm-specific determinants into our model (columns one, three, and five). In the second step, we report the estimates of the full model with the firm and macroeconomic factors (columns two, four, and six).

Empirical results
The Sargan test reveals no evidence of over-identifying restrictions. Our lagged dependent variable, which measures the degree of persistence of our performance measure ROA, is statistically salient during all periods, indicating a high degree of persistence in firm performance and justifying the use of a dynamic model. There are some pronounced differences between the results of the crisis and non-crisis periods, both in respect to the significance and the sign of the coefficients.
The cash flow per share and leverage have positive and significant impacts on firm performance during the non-crisis period, but they do not have a significant effect on firm performance as measured by ROA during the financial crisis. In contrast, the dividend yield ratio, audited by Big Four accounting firms, and audit opinion all have significantly negative impacts on firm performance during the non-crisis period, but do not have a significant effect during the crisis. One possible reason we provide for the above five impact changes from significant in the non-crisis period to insignificant in the crisis period is that during the crisis stock markets became less efficient in incorporating firm-specific information into stock prices (Gupta et al., 2013). Aebi et al. (2012) pinpoint that corporate governance variables are mostly insignificantly related to banks' performance during the crisis, rather than during the non-crisis period. Regarding the negative influence of dividend yield during the non-crisis period, our finding is consistent with Gaver and Gaver (1993) in that growth firms have significantly lower dividend yields than non-growth firms. As for audited by Big Four accounting firms, Mitton (2002) finds that a significantly better performance is associated with firms that had indicators of higher disclosure quality (i.e., auditors from Big Four accounting firms). Conversely, we find that firms audited by Big Four accounting firms have lower performance, implying audit by Big Four accounting firms brings firms a higher degree of quality guarantee or commitment. The result is in line with Liu et al. (2012), whereby Chinese firms that adopted a reputable accounting auditor experienced a small reduction in firm value during the global financial crisis. notes: ***, **, and * indicate the 1, 5, and 10% significant levels, respectively. For the notation of the variables, see table 2.
Data source: https://www.cninfo.com.cn.   The coefficients of firm size and P/E have a negative and highly significant effect on firm performance during both the crisis and non-crisis periods. The smaller the firm size is, the higher its performance is. This result does not meet the disclosure quality expectation, but stands in line with the results of Halkos and Tzeremes (2007) in that small firms could be more efficient, because they have flexible, non-hierarchical structures and do not usually suffer from the so-called agency problem. The P/E has a largely negative impact on firm performance and is in line with Basu (1983) and Jaffe et al. (1989), who find positive abnormal returns associated with stocks having a low P/E. Firm age and the top ten shareholding ratio have significantly positive impacts on firm performance during the crisis period, but salient and negative impacts during the noncrisis period. The negative impact of firm age during the non-crisis period is consistent with Coad et al. (2013) in that firm performance deteriorates with age. However, young firms are more financially constrained (Binks & Ennew, 1996), especially during economic downturn. During the crisis period, an increase in the top ten shareholder ratio brings about higher firm performance, which is in line with Douma, George, and Kabir (2006), who report that a larger foreign ownership shareholding brings a better of firm commitment and long-term involvement. During the non-crisis period, firm performance is lower when large foreign shareholders have the opportunity and the incentive to expropriate minority shareholders (Bozec & Laurin, 2008).

ROA
As for the ownership structure variables, public share ratio and foreign institutions' share ratio have significant and positive impacts on firm performance during the noncrisis period, but they have no salient impact during the crisis period. In contrast, domestic institutions' share ratio and state share ratio have significantly negative impacts on firm performance during the non-crisis period, but they have no salient impact during the crisis period. Our finding is consistent with Liu et al. (2012), who report that Chinese stateowned enterprises performed poorly before the global financial crisis and performed better during the crisis, suggesting that state ownership mitigates financial constraints during the financial crisis. Wei, Xie, and Zhang (2005) also find that state share ratio is significantly negatively related to firm value. In line with Douma et al. (2006) and Wei et al. (2005), we find a significantly positive effect of foreign ownership on firm performance during the non-crisis period. In addition, Mitton (2006) notes that firms having more foreign investors have higher growth and greater profitability.
Considering the macroeconomic factors and firm performance, the consumer confidence index and exports have saliently positive impacts on firm performance for the full period, while M2 and the PPI are negatively correlated with firm performance during the noncrisis period. The positive effect of the consumer confidence index confirms the finding of Qiu and Welch (2005), who support the use of the consumer confidence index as a proxy for sentiment and find that the consumer confidence index correlates well with investors' optimism of future market-wide performance.
The extant research on exports can be divided into two categories: macro level and micro level (Singh, 2009). In micro-level exporting research, Temouri, Vogel, and Wagner (2013) do not find evidence for the positive effects of exporting on firm performance. In macro-level research, we find that exports have a positive effect on firm performance, especially during the non-crisis period. Lorie and Hamilton (1973) find evidence that changes in the money supply rate lead to changes in equity prices. Our finding is in line with Haron (1996) for an adverse effect of money supply (M2) on bank profits. The reason for the negative impact of M2 on firm performance seems to be that a higher M2 indicates a tighter monetary policy and a reduced general availability of capital in capital markets (Yan, Yang, & Jiao, 2010). Tiwari, Mutascu, and Andries (2013) report that the PPI is important for the measurement of inflation. Du (2006) reports that the relationship between stock returns and inflation can be either negative or positive depending on whether the monetary policy is counter-or pro-cyclical. Our finding is confirmed by Kaul (1987) for a positive relationship between stock returns and inflation during the period of the Great Depression when the monetary policy was pro-cyclical and offers evidence of a negative relationship in the post-World War II period when the monetary policy was countercyclical. Table 6 reports the GMM regression results for our second performance measure, ROE. Again, we estimate the model for the entire time period considered and then separately for the two subsamples of the non-crisis and crisis periods. Overall, the findings of the regressions confirm to a large extent the above-mentioned key results. There are also some differences. The current ratio and P/B ratio have no impact on firm performance when measured by the ROA, while the measure of firm performance on ROE has a saliently negative correlation with the current ratio in the crisis period and a positive correlation with P/B in both the full year and the crisis periods. This finding might be driven by the fact that the ROE reveals shareholder maximisation attempts (Dietrich & Wanzenried, 2011). The current ratio is always a proxy for slack resources, and until now a consensus on knowing how slack resources affect performance remains elusive (Daniel, Lohrke, Fornaciari, & Turner, 2004). Our results confirm Davis and Stout (1992) in that a resource slack may become a source of agency problems, especially during turbulent times (Sharfman, Wolf, Chase, & Tansik, 1988). The high correlation of ROE and P/B is consistent with the finding of Bloomfield, Libby, and Nelson (1999) whereby security values are determined by each firm's price/book ratio, and investors are given information on a firm's ROE. Table 7 presents the regression results for another performance measure, the EPS. Analysing the determinants of the EPS helps us to better understand firm profitability. Compared with the results in Tables 5 and 6, the relationships of firm age and the top ten shareholder ratio show a significantly positive impact on firm performance, especially during the crisis period rather than in the non-crisis period. The results of Table 7 confirm to a large extent the key findings of Tables 5 and 6. In other words, we find a negative impact from a worse than unqualified audit opinion during the non-crisis period, but no impact during the crisis; a positive impact of leverage during the non-crisis period; a positive impact of P/B during all periods; and negative impacts of dividend yield and audited by Big Four accounting firms during the non-crisis period. Except for a salient positive impact of cash flow per share during the crisis period and no significant impacts shown four ownership structure variables in both subsample periods, we find strong evidence that all ownership structure and macroeconomic variables are significant impacts in our all-sample period. The possible reason for significant results in all-sample period compared with insignificant results in subsample periods is because the length of data is big enough to capture the potential effects on firm performance.

Discussion
At the point of the global financial crisis hit, why did some firm perform better? Our main findings are as follows. First, firms having higher top ten shareholder ratios or firms that Table 6. Regression results of RoE as dependent variable.
notes: the table reports results from Gmm estimations of the effects of firm-level and macroeconomic-level variables on firm profitability. the dependent variable is the RoE. the estimation method is the two-step Gmm dynamic panel estimator. For the notation of the variables, see table 2. the full sample includes 18,876 observations from 1452 firms. the full period covers the years 1999 to 2010, before the financial crisis 1999 to 2006, and after the financial crisis 2007 to 2010. P-values are in parentheses. coefficients that are significantly different from zero at the 1%, 5%, and 10% levels are marked with ***, **, and *, respectively. the null hypothesis of the sargan test is that the instruments used are not correlated with residuals (over-identifying restrictions). Data source: https://www.cninfo.com.cn.   are older performed poorly during the non-crisis period, yet performed better in the crisis period. Most top ten shareholders of Chinese-listed companies are state government agencies and legal persons (Xu & Wang, 1999). A possible economic insight for this finding is that increases in top ten shareholder ratios stabilise firm operations and older firms are less financially constrained, especially during a crisis period. Our finding is consistent with Liu et al. (2012), who report that Chinese state-owned enterprises performed poorly before the global financial crisis and performed better during the crisis, suggesting that state ownership mitigates financial constraints during the financial crisis. Young firms are more financially constrained (Binks & Ennew, 1996), especially during economic downturn.
For both crisis and non-crisis periods, firm size has pronouncedly negative impacts on firm performance, confirming with Halkos and Tzeremes (2007) that small firms could be more efficient, because they have flexible, non-hierarchical structures and do not usually suffer from the agency problem. Firms audited by the Big Four accounting firm have a significantly negative correlation with their performance, signifying that the strict requirements held up by the Big Four accounting firms might lower firm performance. In other worlds, an individual firm that adopts a reputable accounting auditor experiences a reduction in its value. Leverage and audit opinions that are worse than unqualified audit opinions have salient impacts on firm performance during the non-crisis period rather than during the crisis period, implying that stock markets during a crisis become less efficient in incorporating firm-specific information into stock prices. Considering the macroeconomic factors, the consumer confidence index and macro-level exports have significant positive impacts on firm performance during all years. The positive impact of the consumer confidence index confirms the findings of Qiu and Welch (2005) that the consumer confidence index correlates well with investors' optimism of future market-wide performance.
The subtle economic insights of our analysis are as follows: first, firm performance varies on firm size, history, leverage, audit by the Big Four accounting firms, audit opinion, top ten shareholder ratio, time periods (non-crisis and crisis periods), and on macroeconomic factors (export and consumer confidence index). Second, investors need to be more cautious if they invest in younger companies and firms have a smaller top ten shareholder ratio during the crisis period; while firms with larger leverage and audit opinion are worse than unqualified audit opinions during the non-crisis period; larger firms during the full period. Third, the Chinese financial markets became less efficient with respect to information diffusion during the crisis. It is possible, then, that investors or asset managers who build both leverage and audit opinion that are worse than unqualified audit opinions into investment strategies would find that when it comes to financial crises, their strategies may not deliver the desired investment outcome. Our findings indicate that policymakers should consider the key factors that fuel their economy, and firm managers should effectively identify and adopt those characteristics that cause their firm to perform better during different periods and under different macroeconomic conditions. The main contribution of the paper is applicable to China, and is broadly generalisable to other developing market contexts.

Conclusions
Contrary to most recent literature which emphasises the issues of bank performance and/or corporate governance during the recent global financial crisis, this paper captures integrated rather than segmented factors, combining firm, macroeconomic, and different economic environments for the purpose of firm performance analysis. Given that there are wide variances in firms' capabilities as well as countries' economic policy, an imperative issue is why some firms perform better and others suffer during a drastic financial environmental change. This paper examines the impacts of firm and macroeconomic factors on firm performance during the recent global crisis and during the non-crisis period, with a wide range of panel data that cover 1,452 Chinese public non-financial firms from the period 1999 to 2010.
In this research we consider three categories of firm variables (i.e., financial structure, disclosure quality, and ownership structure) and macroeconomic variables during crisis and non-crisis periods. This study applies the recent two-step system GMM dynamic panel data techniques. The empirical findings reveal that the effects of firm age and the top ten shareholder ratio on firm performance are significantly positive (negative) during the crisis (non-crisis) period, which is consistent with the findings of Binks and Ennew (1996), meaning young firms are more financially constrained on their performance, and in line with Douma et al. (2006) that a larger foreign ownership shareholding brings a better firm commitment and long-term involvement during the crisis period. The negative impacts of firm size and P/E during both the non-crisis and crisis periods confirm the results of Halkos and Tzeremes (2007) in that small firms may more efficient due to the agency problem, and agree with Basu (1983), who state that P/E presents positive abnormal returns associated with stocks having a low P/E.
Firms audited by the Big Four accounting firms are significantly negatively correlated with firm performance, signifying that the strict requirements upheld by the Big Four accounting firms might lower firm performance (net income). Firms that adopted Big Four accounting auditors experienced a small reduction in firm value, confirming the results of Liu et al. (2012). Cash flow per share, dividend yield ratio, and leverage have a salient impact on ROA and ROE during the non-crisis period, while they have no significant impact during the crisis period. This is consistent with Gupta et al. (2013), who suggest that during the crisis stock markets became less efficient in incorporating firm-specific information into stock prices. Considering the macroeconomic factors, the consumer confidence index and macro-level exports have significant impacts on firm performance during all years, indicating the strength of the consumer confidence index and exports on firm performance. The negative effect of money supply (M2) on ROA and ROE is in line with Haron (1996), who shows an adverse effect of M2 on profits, implying a higher M2 tightens a monetary policy and reduces the general availability of capital in capital markets (Yan et al., 2010).
The contributions of this study are as follows. First, this study accounts for the contingency variables of the macroeconomy. Second, the extant literature ignores influencing factors that might change under different market conditions. Third, this study employs a GMM approach to illustrate the wealth effect of the selected factors. Our results provide some important policy implications for policy makers, firm managers, and investors during crisis and non-crisis periods. During non-crisis and crisis periods, policy makers may attempt to avert a financial crisis or improve the economy by giving priority to accelerating the consumer confidence index and exports as well as decrease money supply. As to firm managers, they can strengthen the ownership structure of a firm during a crisis. More precisely, pushing the top ten shareholders to increase their shareholdings and considering key performance factors (i.e., P/E) should help maximise firm performance during a crisis. During a non-crisis period, firms should put more weight on firms' cash flow per share, dividend yield, and leverage in order to strengthen their firm's quality as well as create performance. Due to the different impacts of financial factors (e.g., cash flow per share, dividend yield, and leverage) in crisis and non-crisis periods, our results suggest that investors should be more cautious in basing their investment decisions on the informativeness of financial factors during crisis, and give more weight to top ten shareholder ratio, firm history, and firm size to estimate risks, since these three factors are harder to manipulate. Furthermore, while firm managers aim at critical factors that could saliently improve firm performance, economic growth would accelerate more if matched with effective government economic policy.
Financial crises do repeat (Reinhart & Rogoff, 2009), and appropriate responses to them are difficult and complex. To be better prepared to face the next financial crisis, we conclude that firms have to significantly improve their disclosure quality and financial and ownership structures. The government should set up appropriate monetary policy and economic governance, so that the next crisis will not produce harmful consequences. The findings herein based on the experience in China provide important insights for firms and countries to follow. Evidence provided in this study on the ability to distinguish better firms from other firms ex-ante a crisis adds to the literature on understanding corporate and macroeconomic features, and is relevant to decisions that enhance the effectiveness of policymakers on country governance as well as management on a firm's operations. Finally, investors can identify the most common variables used in our analyses for their own beneficial use.

Notes
1. There are five types of audit opinions: unqualified, unqualified opinion with emphasis of matter, qualified, disclaimer, and adverse. 2. Listings in the U.S., which require more stringent regulations and disclosure rules, may strengthen firms' governance practices and thereby enhance firm value (Pan, Lin, & Chen, 2012). As both Level II and III ADRs are required to follow more stringent requirements on governance and accounting standards versus U.S. firms (Durnev & Kim, 2005;Mitton, 2002), this study uses the dummy variable if firms have Level II and Level III ADRs equal to 1. However, among the 1,452 sample firms, only 12 firms have issued Level II or III ADRs, resulting in a single matrix of regression finding. Thus, we remove this disclosure quality dummy variable. 3. Source: Producer price index: Methods, sources and theory. www.statssa.gov.za. 4. We do consider macroeconomic factors, such as GDP growth rate, unemployment rate, reserve rate, exchange rate to U.S. dollars, government expenditure, business cycle signal, foreign direct investment, and real interest rate, but remove them due to the large correlations among them. 5. We consider industry effect, exchange market effect (Shanghai and Shenzhen), A-share, or B-share, but no significant influences are found. 6. The OLS approach does not account formally for potential simultaneity bias, or does it control explicitly for firm fixed effects or the routine use of lagged dependent variables in regressions. Unlike the OLS approach, the GMM panel estimator exploits the time-series variation in the data, accounts for unobserved firm-specific effects, allows for the inclusion of lagged dependent variables as regressors, and controls for endogeneity of all the explanatory variables. Thus, this study advances the literature on firm performance, firm, and macroeconomic factors by enhancing the quality and quantity of the data and by using econometric techniques that ameliorate biases. 7. Use a GMM estimator also accounts for possible correlations between any of the independent variables. For a thorough description of the various GMM estimators, see Baltagi (2001) 8. Mitton (2002) presents that the Asian financial crisis is from 1997. 9. Choi (2013 Table 3 lists the 11 industries used in this study.